by G. Julián Cervera Leonetti
For my final project in the graduate-level course PA 5929: Data Visualization, I employed Python's plotly express library to create two interactive graphs. The objective of this exercise was to select a dataset that held personal relevance, generate two distinct visualizations illustrating the same aspect of the data, determine the superior visualization style, and articulate the reasoning behind the choice. The dataset chosen for this project comprises sales reports from Guatemalan forestry concessions operating within the Maya Biosphere Reserve in the province of Petén, Guatemala. Recognized as one of the most significant rainforests globally, these reports play a pivotal role in comprehending the sustainability achievements and challenges faced by the reserve. Personally, this data holds particular significance as it was collected by the Rainforest Alliance, the non-governmental organization for which I served as a graduate student consultant during the summer of 2023. This project can be viewed as a natural extension of the exploratory data analysis conducted for the organization during that period. The visualizations presented here focus on depicting the total sales of all forestry concessions from 2018 to 2022, quarter by quarter. It is important to note that data for the first quarter of 2018 and the last quarter of 2022 are missing, as they were not included in the dataset. The code behind the graphs can be found and independently run here.
I opted for two distinct techniques to effectively communicate the data. The first method employs an interactive line graph, which proves highly effective in portraying the sequential ebbs and flows in the data, quarter by quarter. This visualization method captures the subtleties of fluctuations over time, providing a nuanced understanding of sales trends. Notably, it accentuates quarters characterized by exceptionally high sales, a detail that might be overlooked if solely comparing years. The interactivity of this graph allows users to hover over the line, obtaining precise information about the sales figures for each quarter. Additionally, users can click to zoom in on specific periods of interest, offering a more detailed examination, for instance, to focus on sales patterns within a particular year.
Nevertheless, my preferred visualization technique for this dataset is the stacked bar chart. Each bar in the chart represents a year, facilitating straightforward year-by-year comparisons. The interactivity feature involves clicking on the legend to selectively hide certain quarters, allowing a focused view of every year's first quarter, second quarter, and so forth. Additionally, hovering the cursor over the segments within each bar provides detailed information on total sales per quarter.
The stacked bar chart proves particularly useful for comparing sales on both an annual and quarterly basis. This aligns with economic cycles tied to fiscal budgeting and taxation, providing insights into the yearly variation of sales performance. Notably, the chart enables users to compare sales for corresponding trimesters across different years (e.g., 2019 Q1 versus 2020 Q1), shedding light on seasonal trends. Given that the products in consideration are primarily agricultural and subject to seasonal cycles, understanding these variations is crucial. However, it's worth noting that, despite its advantages, the stacked bar chart may obscure peak sales moments, as exemplified by the best-selling period in 2018's fourth quarter. Although this quarter had the highest sales, its bar appears shorter compared to other years, making the peak less conspicuous at first glance, unlike the line graph.
In conclusion, it is evident that both graphs possess unique strengths and weaknesses. The line graph excels in illustrating the flow of sales over time, providing a comprehensive view of sequential trends. On the other hand, the stacked bar chart offers the advantage of allowing users to aggregate or disaggregate time periods as required for detailed comparisons. Each visualization method caters to specific analytical needs, and the choice between them ultimately depends on the specific insights sought and the nature of the data being presented.
The paragraphs above synthesize information from original bullet points provided by me to ChatGPT (OpenAI, 2023), which was prompted to transform the bullet points into a cohesive text, using only the information provided by me. OpenAI. (2023). ChatGPT (Dec 8 Version) [Large language model]. https://chat.openai.com/chat